|
| Work | Type | Sensors | Features | Comments | Results | Algorithms |
|
| [58] | Motion | Gyroscope and touchscreen | Time domain, angle of upper bisector, and angle of lower bisector | Digit-only soft keyboard | 70% accuracy for a 4-digit PIN | Guess classifier |
| TapLogger [197] | Motion | Gyroscope, accelerometer, and touchscreen | Time domain and angle changes | Soft keyboard for digits | 90% accuracy with 3 traits for an 8-digit PIN | SVM using LIBSVM [198] |
| TapPrints [183] | Motion | Gyroscope and Accelerometer | Time domain, frequency domain, and FTT values | Soft keyboard for English characters | 90% accuracy for English character inferring | -nearest neighbor (KNN), multinomial logistic regression, SVM, random forests [199], and bagged decision trees |
| Accessory [184] | Motion | Accelerometer | Time domain, the average time from a sample to a peak, the total time of the window, and the number of samples in the window | Soft keyboard for English characters | 6 password characters in 4.5 trails | Random forest, ANN, SVM, and C4.5 decision tree |
| [200] | Acoustic | Microphone | Cepstrum features [201] and speech recognition, claim that it is better than FFT | Soft keyboard for English characters | 96% accuracy | Hidden Markov models [202], linear classification, ANN, and language models have been used |
| Timing attack [142] | Timing | Microphone | FFT | Soft keyboard and hard keypad | Inferring PIN without triggering alerts | Markov chainswith brute force attack |
| Soundminer [203] | Acoustic | Microphone | FFT, voice record, and dual-tone multifrequency (DTMF) | Soft keyboard | Inferring PIN, passwords, volume up and down keys | Google speech recognition |
| Powerspy [145] | Power | Power usage of wireless communication | Power usage files as time series | — | Inferring routes and real-time tracking dynamic | Time warping (DTW) [204] and optimal subsequence bijection (OSB) [205] |
| [149] | Power | Power usage of the smartphone | Power usage files as time series | — | Inferring apps, geolocation, password length, and UI | Statistics |
| [153] | Timing | Interrupts | Interrupt file | — | Inferring apps and unlocking the phone | DWT and hidden Markov model (HMM) |
| PinMe [151] | Motion | Accelerometer, gyroscope, barometer, IP, and time zone | Sensor data | — | Inferring and tracking users around the world | SVM |
| [157] | Acoustic | Microphone | Row data | — | Accuracy of more than 70% for 5-minute audio files | Statistical |
| [156] | Acoustic | Microphone | Row data | — | 1.5 error rate | Unsupervised algorithm |
|
